[PATCH BlueZ 1/3] attrib: Fix not honoring MTU

[PATCH BlueZ 1/3] attrib: Fix not honoring MTU

[Luiz Augusto von Dentz]

From: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>

If MTU has changed, by bt_gatt_client for example, the code should be
able to figure it out without waiting g_attrib_set_mtu to be called.
---
 attrib/gattrib.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/attrib/gattrib.c b/attrib/gattrib.c
index 514fc8f..2011359 100644
--- a/attrib/gattrib.c
+++ b/attrib/gattrib.c
@@ -429,9 +429,24 @@ guint g_attrib_register(GAttrib *attrib, guint8 opcode, guint16 handle,
 
 uint8_t *g_attrib_get_buffer(GAttrib *attrib, size_t *len)
 {
+	uint16_t mtu;
+
 	if (!attrib || !len)
 		return NULL;
 
+	mtu = bt_att_get_mtu(attrib->att);
+
+	/*
+	 * Clients of this expect a buffer to use.
+	 *
+	 * Pdu encoding in shared/att verifies if whole buffer fits the mtu,
+	 * thus we should set the buflen also when mtu is reduced. But we
+	 * need to reallocate the buffer only if mtu is larger.
+	 */
+	if (mtu > attrib->buflen)
+		attrib->buf = g_realloc(attrib->buf, mtu);
+
+	attrib->buflen = mtu;
 	*len = attrib->buflen;
 	return attrib->buf;
 }
-- 
2.1.0

[PATCH BlueZ 2/3] core/device: Fix not handling notification

[Luiz Augusto von Dentz]

From: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>

attio callbacks needs to be triggered as soon as possible once connected
otherwise profiles such as HoG may miss notification that are sent while
bt_gatt_client is doing MTU exchange.
---
 src/device.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/src/device.c b/src/device.c
index 831efea..0dc8c86 100644
--- a/src/device.c
+++ b/src/device.c
@@ -3984,8 +3984,6 @@ static void gatt_client_ready_cb(bool success, uint8_t att_ecode,
 
 	device_accept_gatt_profiles(device);
 
-	g_slist_foreach(device->attios, attio_connected, device->attrib);
-
 	btd_gatt_client_ready(device->client_dbus);
 }
 
@@ -4007,6 +4005,9 @@ static void gatt_client_init(struct btd_device *device)
 		return;
 	}
 
+	/* Notify attio so it can react to notifications */
+	g_slist_foreach(device->attios, attio_connected, device->attrib);
+
 	if (!bt_gatt_client_set_ready_handler(device->client,
 							gatt_client_ready_cb,
 							device, NULL)) {
-- 
2.1.0

[PATCH BlueZ 3/3] core/device: Remove call to g_attrib_set_mtu

[Luiz Augusto von Dentz]

From: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>

The call to g_attrib_set_mtu is no longer needed since
g_attrib_get_buffer has learned to check bt_att MTU.
---
 src/device.c | 5 -----
 1 file changed, 5 deletions(-)

diff --git a/src/device.c b/src/device.c
index 0dc8c86..97c0523 100644
--- a/src/device.c
+++ b/src/device.c
@@ -3975,11 +3975,6 @@ static void gatt_client_ready_cb(bool success, uint8_t att_ecode,
 		return;
 	}
 
-	device->att_mtu = bt_att_get_mtu(device->att);
-	g_attrib_set_mtu(device->attrib, device->att_mtu);
-
-	DBG("MTU: %u", device->att_mtu);
-
 	register_gatt_services(device);
 
 	device_accept_gatt_profiles(device);
-- 
2.1.0

Re: [PATCH BlueZ 1/3] attrib: Fix not honoring MTU

[Luiz Augusto von Dentz]

Hi,

On Fri, Mar 13, 2015 at 10:54 AM, Luiz Augusto von Dentz
<luiz.dentz@gmail.com> wrote:
> From: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
>
> If MTU has changed, by bt_gatt_client for example, the code should be
> able to figure it out without waiting g_attrib_set_mtu to be called.
> ---
>  attrib/gattrib.c | 15 +++++++++++++++
>  1 file changed, 15 insertions(+)
>
> diff --git a/attrib/gattrib.c b/attrib/gattrib.c
> index 514fc8f..2011359 100644
> --- a/attrib/gattrib.c
> +++ b/attrib/gattrib.c
> @@ -429,9 +429,24 @@ guint g_attrib_register(GAttrib *attrib, guint8 opcode, guint16 handle,
>
>  uint8_t *g_attrib_get_buffer(GAttrib *attrib, size_t *len)
>  {
> +       uint16_t mtu;
> +
>         if (!attrib || !len)
>                 return NULL;
>
> +       mtu = bt_att_get_mtu(attrib->att);
> +
> +       /*
> +        * Clients of this expect a buffer to use.
> +        *
> +        * Pdu encoding in shared/att verifies if whole buffer fits the mtu,
> +        * thus we should set the buflen also when mtu is reduced. But we
> +        * need to reallocate the buffer only if mtu is larger.
> +        */
> +       if (mtu > attrib->buflen)
> +               attrib->buf = g_realloc(attrib->buf, mtu);
> +
> +       attrib->buflen = mtu;
>         *len = attrib->buflen;
>         return attrib->buf;
>  }
> --
> 2.1.0

Applied.


-- 
Luiz Augusto von Dentz